Nothing Special   »   [go: up one dir, main page]

EP0885422A1 - Process for classification of a time series which contains a pre-set number of sample values, particularly of an electrical signal, by means of a computer - Google Patents

Process for classification of a time series which contains a pre-set number of sample values, particularly of an electrical signal, by means of a computer

Info

Publication number
EP0885422A1
EP0885422A1 EP97914158A EP97914158A EP0885422A1 EP 0885422 A1 EP0885422 A1 EP 0885422A1 EP 97914158 A EP97914158 A EP 97914158A EP 97914158 A EP97914158 A EP 97914158A EP 0885422 A1 EP0885422 A1 EP 0885422A1
Authority
EP
European Patent Office
Prior art keywords
time series
classification
signal
computer
sample values
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
EP97914158A
Other languages
German (de)
French (fr)
Inventor
Gustavo Deco
Bernd SCHÜRMANN
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Siemens AG
Original Assignee
Siemens AG
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Siemens AG filed Critical Siemens AG
Publication of EP0885422A1 publication Critical patent/EP0885422A1/en
Withdrawn legal-status Critical Current

Links

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A61B5/349—Detecting specific parameters of the electrocardiograph cycle
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/369—Electroencephalography [EEG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
    • A61B5/1455—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue using optical sensors, e.g. spectral photometrical oximeters
    • A61B5/14551—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue using optical sensors, e.g. spectral photometrical oximeters for measuring blood gases
    • A61B5/14553—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue using optical sensors, e.g. spectral photometrical oximeters for measuring blood gases specially adapted for cerebral tissue
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems

Definitions

  • the invention relates to technical fields in which it is of interest to infer the future behavior of the time series from measured time series.
  • This prediction of the future “behavior” of the time series takes place on the assumption that the time series has non-linear correlations or linear correlations (statistical dependencies) between the samples of the time series.
  • This problem is also of considerable importance in various areas of medicine, for example in cardiology. Particularly in the problem area of sudden cardiac death, it can be vital to recognize early warning signs of sudden cardiac death in order to take countermeasures against the occurrence of sudden cardiac death as early as possible.
  • a time series of an electrocardiogram which is not correlated, describes a heart that is not at risk with regard to sudden cardiac death.
  • a vulnerable heart with regard to sudden cardiac death is described by a time series of the electrocardiogram, which has non-linear correlations between the samples of the time series [1].
  • [1] it is known from [1] to determine time series of an electrocardiogram from the graphical phase space representation of two successive heartbeats, which describe hearts which are at risk of sudden cardiac death.
  • a method is known from [2] with which the time course of the local oxygen tension in the brain (tip02) can be determined.
  • the method described in [1] has all the disadvantages inherent in empirical methods.
  • the susceptibility of graphic interpretations by a human being, the problem of setting a barrier from which a time series is classified as at risk, and inaccuracies in the representation of the Fourier transforms on the screen are to be regarded as disadvantages of the known method.
  • the invention is based on the problem of classifying a time series analytically with the aid of a computer.
  • a parameterized dynamic diversity is determined for a time series which has a predeterminable number of samples.
  • the parameterized dynamic manifold describes nonlinear correlations between the samples of the time series by parameters that indicate the form of the parameterized dynamic manifold.
  • the time series is classified based on the parameterized dynamic diversity.
  • the method Compared to the known empirical method, the method has the particular advantage that no sources of error due to inaccurate evaluation of the results and no inaccurate results are determined.
  • the analytical procedure achieves a clear, understandable classification of the time series. Another advantage can be seen in the significantly increased speed with which the entire classification process is carried out.
  • a first time series type describes a time series in which a nonlinear correlation between the samples is determined and the second time series type a time series that is statistically independent.
  • the parameterized dynamic diversity differs so much that a classification can be carried out very quickly, without a great deal of computation.
  • Figure 1 is a flowchart describing the method steps of the invention
  • FIG. 2 shows a flowchart which describes the development of the method according to the invention with a classification into only two time series types
  • FIG. 3 shows a block diagram in which various possibilities are shown, of what type the time series can be, for example
  • FIG. 4 shows a sketch in which a computer with which the method is carried out is shown.
  • a time series is measured if the time series is an electrical signal acts. Since the time series is classified using a computer R, the time series has a predeterminable number of sample values, depending on a sample interval. If the measured signal, which represents the time series, is analog, the signal must be sampled so that it can be processed with the computer R.
  • a parameterized dynamic diversity is determined for the time series.
  • the parameterized dynamic manifold is ascertained in such a way that nonlinear correlations which the sample values have among one another are extracted.
  • a method for determining the parameterized dynamic diversity is known [3].
  • a last step 103 the time series is classified on the basis of the parameterized dynamic diversity.
  • the method is particularly suitable for use in medical areas.
  • the early detection of sudden cardiac death on the basis of electrocardiogram signals (EKG signals) is an important area of application for the method according to the invention.
  • Hearts at risk of sudden cardiac death in the electrocardiogram signals are characterized by the presence of nonlinear correlations between the samples of the electrocardiogram signal.
  • Hearts not at risk with regard to sudden cardiac death have no correlations between the samples of the time series in the EKG signal, the samples are statistically independent of one another.
  • FIG. 2 shows a further development of the method according to the invention, which considerably accelerates the feasibility of the method from measuring the time series to classifying the time series. This is achieved through a simplified classification.
  • the simplification is that only on the basis of the parameterized dynamic diversity is it checked whether the samples of the time series are statistically dependent on one another 201. If this is the case, the time series is classified 202 as a first time series type 202.
  • This classification corresponds for the application example of the classification of electrocardiogram signals in endangered and non-endangered hearts with regard to the sudden cardiac death of a classification of the electrocardiogram signal as a signal of an endangered heart, since in this case non-linear correlations between the samples of the time series, in in this case of the electrocardiogram signal.
  • the time series is classified 203 as a second time series type.
  • this corresponds to a classification as a signal of a heart that is not at risk with regard to sudden cardiac death.
  • Electrocardiogram signals EKG signals
  • Electroencephalogram signals EEG signals
  • the method can of course be used in all areas in which it is important to classify a time series on the basis of parameterized dynamic manifolds.
  • FIG. 4 shows the computer R with which the method according to the invention is necessarily carried out.
  • the computer R processes the time series recorded by the measuring device MG and fed to the computer R.
  • the measuring device MG can be, for example, an electrocardiograph (EKG), an electroencepahlograph (EEG) or also a device which works according to the method shown in [2].
  • EKG electrocardiograph
  • EEG electroencepahlograph
  • the classification result which is ascertained by the computer R in the manner described above, is further processed in a means for further processing WV, for example presented to a user.
  • This means WV can be, for example, a printer, a screen or a loudspeaker, via which an acoustic or visual signal is passed on to a user.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Biophysics (AREA)
  • General Health & Medical Sciences (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Molecular Biology (AREA)
  • Medical Informatics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Psychiatry (AREA)
  • Cardiology (AREA)
  • Evolutionary Computation (AREA)
  • Signal Processing (AREA)
  • Physiology (AREA)
  • Mathematical Physics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Fuzzy Systems (AREA)
  • Psychology (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

A parameterized, dynamic multiplicity is determined for a time series and used to describe non-linear correlations between sample values in the time series. The time series is classified by means of the parameterized dynamic multiplicity.

Description

Beschreibungdescription
Verfahren zur Klassifikation einer Zeitreihe, die eine vor- gebbare Anzahl von Abtastwerten aufweist, insbesondere eines elektrischen Signals, durch einen RechnerMethod for classifying a time series which has a predeterminable number of samples, in particular an electrical signal, by a computer
Die Erfindung betrifft technische Gebiete, in denen es von Interesse ist, aus gemessenen Zeitreihen auf das zukünftige Verhalten der Zeitreihen zu schließen. Diese Vorhersage des zukünftigen „Verhaltens" der Zeitreihe erfolgt unter der An¬ nahme, daß die Zeitreihe nichtlineare Korrelationen oder li¬ neare Korrelationen (statistische Abhängigkeiten) zwischen den Abtastwerten der Zeitreihe aufweist.The invention relates to technical fields in which it is of interest to infer the future behavior of the time series from measured time series. This prediction of the future “behavior” of the time series takes place on the assumption that the time series has non-linear correlations or linear correlations (statistical dependencies) between the samples of the time series.
Erhebliche Bedeutung erlangt dieses Problem auch in verschie¬ denen Gebieten der Medizin, beispielsweise in der Kardiolo- gie. Speziell in dem Problembereich des Plötzlichen Herztodes kann es lebenswichtig sein, Frühwarnzeichen des Plötzlichen Herztodes zu erkennen, um so früh wie möglich Gegenmaßnahmen gegen das Eintreten des Plötzlichen Herztodes einzuleiten.This problem is also of considerable importance in various areas of medicine, for example in cardiology. Particularly in the problem area of sudden cardiac death, it can be vital to recognize early warning signs of sudden cardiac death in order to take countermeasures against the occurrence of sudden cardiac death as early as possible.
Es ist bekannt, daß eine Zeitreihe eines Elektrokardiogramms, welches nicht korreliert ist, ein nicht gefährdetes Herz be- züglich des plötzlichen Herztodes beschreibt. Ein gefährdetes Herz bezüglich des plötzlichen Herztodes wird durch eine Zeitreihe des Elektrokardiogramms beschrieben, welches nicht- lineare Korrelationen zwischen den Abtastwerten der Zeitreihe aufweist [1] . Weiterhin ist es aus [1] bekannt, aus der gra- phischen Phasenraumdarstellung zweier aufeinanderfolgender Herzschläge Zeitreihen eines Elektrokardiogramms zu ermit¬ teln, die Herzen beschreiben, die bezüglich des Plötzlichen Herztodes gefährdet sind.It is known that a time series of an electrocardiogram, which is not correlated, describes a heart that is not at risk with regard to sudden cardiac death. A vulnerable heart with regard to sudden cardiac death is described by a time series of the electrocardiogram, which has non-linear correlations between the samples of the time series [1]. Furthermore, it is known from [1] to determine time series of an electrocardiogram from the graphical phase space representation of two successive heartbeats, which describe hearts which are at risk of sudden cardiac death.
Aus [2] ist ein Verfahren bekannt, mit dem der zeitliche Ver¬ lauf der lokalen SäuerstoffSpannung des Gehirns (tip02) er¬ mittelt werden kann. Das in [1] beschriebene Verfahren weist alle Nachteile auf, die empirische Verfahren in sich birgen. Hierbei sind insbe¬ sondere die Fehleranfälligkeit graphischer Deutungen durch einen Menschen, das Problem des Setzens einer Schranke, ab der eine Zeitreihe als gefährdet klassifiziert wird, sowie Ungenauigkeiten in der Darstellung der Fourier- Transformierten auf dem Bildschirm als Nachteil des bekannten Verfahrens zu betrachten.A method is known from [2] with which the time course of the local oxygen tension in the brain (tip02) can be determined. The method described in [1] has all the disadvantages inherent in empirical methods. In particular, the susceptibility of graphic interpretations by a human being, the problem of setting a barrier from which a time series is classified as at risk, and inaccuracies in the representation of the Fourier transforms on the screen are to be regarded as disadvantages of the known method.
Der Erfindung liegt das Problem zugrunde, eine Zeitreihe ana¬ lytisch mit Hilfe eines Rechners zu klassifizieren.The invention is based on the problem of classifying a time series analytically with the aid of a computer.
Das Problem wird durch das Verfahren gemäß Patentanspruch 1 gelöst.The problem is solved by the method according to claim 1.
Bei dem erfindungsgemäßen Verfahren wird für eine Zeitreihe, die eine vorgebbare Anzahl von Abtastwerten aufweist, eine parametrisierte dynamische Mannigfaltigkeit ermittelt. Mit der parametrisierten dynamischen Mannigfaltigkeit werden nichtlineare Korrelationen zwischen den Abtastwerten der Zeitreihe durch Parameter, die die Form der parametrisierten dynamischen Mannigfaltigkeit angeben, beschrieben. Anhand der parametrisierten dynamischen Mannigfaltigkeit wird eine Klas- sifikation der Zeitreihe durchgeführt.In the method according to the invention, a parameterized dynamic diversity is determined for a time series which has a predeterminable number of samples. The parameterized dynamic manifold describes nonlinear correlations between the samples of the time series by parameters that indicate the form of the parameterized dynamic manifold. The time series is classified based on the parameterized dynamic diversity.
Das Verfahren birgt gegenüber dem bekannten empirischen Ver¬ fahren vor allem den Vorteil insich, daß keine Fehlerquellen durch ungenaue Bewertung der Ergebnisse sowie keine ungenauen Ergebnisse ermittelt werden. Durch das analytische Vorgehen wird eine eindeutige, nachvollziehbare Klassifikation der Zeitreihe erreicht. Ein weiterer Vorteil ist in der wesent¬ lich erhöhten Geschwindigkeit zu sehen, mit der das gesamte Verfahren der Klassifikation durchgeführt wird.Compared to the known empirical method, the method has the particular advantage that no sources of error due to inaccurate evaluation of the results and no inaccurate results are determined. The analytical procedure achieves a clear, understandable classification of the time series. Another advantage can be seen in the significantly increased speed with which the entire classification process is carried out.
Vorteilhafte Weiterbildungen der Erfindung ergeben sich aus den abhängigen Ansprüchen. Zur Beschleunigung des Verfahrens ist es vorteilhaft, bei der Klassifikation die Zeitreihe lediglich in einen ersten Zeit¬ reihentyp und in einen zweiten Zeitreihentyp einzuordnen. Hierbei beschreibt ein erster Zeitreihentyp eine Zeitreihe, bei der eine nichtlineare Korrelation zwischen den Abtastwer¬ ten festgestellt wird und der zweite Zeitreihentyp eine Zeit- reihe, die statistisch unabhängig ist. Für diese beiden Zeit¬ reihentypen unterscheidet sich die parametrisierte dynamische Mannigfaltigkeit so stark, daß eine Klassifikation sehr schnell, ohne größeren Rechenaufwand, durchgeführt werden kann.Advantageous developments of the invention result from the dependent claims. To speed up the method, it is advantageous to classify the time series only into a first time series type and into a second time series type. Here, a first time series type describes a time series in which a nonlinear correlation between the samples is determined and the second time series type a time series that is statistically independent. For these two types of time series, the parameterized dynamic diversity differs so much that a classification can be carried out very quickly, without a great deal of computation.
Die Erfindung wird anhand eines Ausführungsbeispiels, welches in den Figuren dargestellt ist, weiter erläutert.The invention is further explained on the basis of an exemplary embodiment which is shown in the figures.
Es zeigenShow it
Figur 1 ein Ablaufdiagramm, welches die Verfahrensschritte der Erfindung beschreibt;Figure 1 is a flowchart describing the method steps of the invention;
Figur 2 ein Ablaufdiagramm, welches die Weiterbildung des erfindungsgemäßen Verfahrens mit einer Klassifika¬ tion in nur zwei Zeitreihentypen, beschreibt;FIG. 2 shows a flowchart which describes the development of the method according to the invention with a classification into only two time series types;
Figur 3 ein Blockdiagramm, in dem verschiedene Möglich¬ keiten dargestellt sind, von welcher Art die Zeit- reihe beispielsweise sein kann;FIG. 3 shows a block diagram in which various possibilities are shown, of what type the time series can be, for example;
Figur 4 eine Skizze, in der ein Rechner, mit dem das Ver¬ fahren durchgeführt wird, dargestellt ist.FIG. 4 shows a sketch in which a computer with which the method is carried out is shown.
In Figur 1 sind die Verfahrensschritte des erfindungsgemäßen Verfahrens dargestellt.The method steps of the method according to the invention are shown in FIG.
In einem ersten Schritt 101 wird eine Zeitreihe gemessen, falls es sich bei der Zeitreihe um ein elektrisches Signal handelt. Da die Klassifikation der Zeitreihe mit einem Rech¬ ner R durchgeführt wird, weist die Zeitreihe eine vorgebbare Anzahl von Abtastwerten, abhängig von einem Abtastintervall, auf. Ist das gemessene Signal, welches die Zeitreihe dar- stellt, analog, so muß das Signal abgetastet werden, damit es mit dem Rechner R verarbeitet werden kann.In a first step 101, a time series is measured if the time series is an electrical signal acts. Since the time series is classified using a computer R, the time series has a predeterminable number of sample values, depending on a sample interval. If the measured signal, which represents the time series, is analog, the signal must be sampled so that it can be processed with the computer R.
Für die Zeitreihe wird in einem zweiten Schritt 102 eine pa- rametrisierte dynamische Mannigfaltigkeit ermittelt. Die Er- mittlung der parametrisierten dynamischen Mannigfaltigkeit erfolgt in einer Weise, daß dabei nichtlineare Korrelationen, die die Abtastwerte untereinander aufweisen, extrahiert wer¬ den. Ein Verfahren zur Ermittlung der parametrisierten dyna¬ mischen Mannigfaltigkeit ist bekannt [3] .In a second step 102, a parameterized dynamic diversity is determined for the time series. The parameterized dynamic manifold is ascertained in such a way that nonlinear correlations which the sample values have among one another are extracted. A method for determining the parameterized dynamic diversity is known [3].
Anhand dieser parametrisierten dynamischen Mannigfaltigkeit ist es nunmehr möglich, Aussagen über eventuell vorhandene nichtlineare Korrelationen zwischen den Abtastwerten zu tref¬ fen.On the basis of this parameterized dynamic manifold, it is now possible to make statements about any nonlinear correlations that may exist between the sample values.
In einem letzten Schritt 103 wird die Zeitreihe anhand der parametrisierten dynamischen Mannigfaltigkeit klassifiziert.In a last step 103, the time series is classified on the basis of the parameterized dynamic diversity.
Das Verfahren eignet sich besonders zur Verwendung in medizi- nischen Bereichen. In diesen Bereichen ist insbesondere die Früherkennung des plötzlichen Herztodes anhand von Elektro¬ kardiogramm-Signalen (EKG-Signale) ein wichtiges Einsatzge¬ biet für das erfindungsgemäße Verfahren.The method is particularly suitable for use in medical areas. In these areas, the early detection of sudden cardiac death on the basis of electrocardiogram signals (EKG signals) is an important area of application for the method according to the invention.
Wie in [1] beschrieben, kann man anhand von Elektrokardio¬ gramm-Signalen feststellen, ob das Herz, für das das Elektro¬ kardiogramm gemessen wurde, bezüglich des Plötzlichen Herzto¬ des gefährdet ist. Hierbei sind bezüglich des Plötzlichen Herztodes gefährdete Herzen in dem Elektrokardiogramm- Signalen durch das Vorhandensein nichtlinearer Korrelationen zwischen den Abtastwerten des Elektrokardiogramm-Signals cha¬ rakterisiert. Bezüglich des plötzlichen Herztodes nicht gefährdete Herzen weisen in dem EKG-Signal keinerlei Korrelationen zwischen den Abtastwerten der Zeitreihe auf, die Abtastwerte sind stati- stisch voneinander unabhängig.As described in [1], it can be determined on the basis of electrocardiogram signals whether the heart for which the electrocardiogram was measured is at risk with regard to the sudden cardiac death. Hearts at risk of sudden cardiac death in the electrocardiogram signals are characterized by the presence of nonlinear correlations between the samples of the electrocardiogram signal. Hearts not at risk with regard to sudden cardiac death have no correlations between the samples of the time series in the EKG signal, the samples are statistically independent of one another.
In Figur 2 ist eine Weiterbildung des erfindungsgemäßen Ver¬ fahrens dargestellt, wodurch die Durchführbarkeit des Verfah¬ rens von der Messung der Zeitreihe bis zur Klassifikation der Zeitreihe erheblich beschleunigt wird. Dies wird erreicht durch eine vereinfachte Klassifikation. Die Vereinfachung be¬ steht darin, daß lediglich anhand der parametrisierten dyna¬ mischen Mannigfaltigkeit überprüft wird, ob die Abtastwerte der Zeitreihe voneinander statistisch abhängig sind 201. Ist dies der Fall, wird die Zeitreihe als ein erster Zeitreihen¬ typ klassifiziert 202. Diese Klassifikation entspricht für das Anwendungsbeispiel der Klassifikation von Elektrokardio¬ gramm-Signalen in gefährdete und nichtgefährdete Herzen be¬ züglich des Plötzlichen Herztodes einer Klassifikation des Elektrokardiogramm-Signals als ein Signal eines gefährdeten Herzens, da in diesem Fall nichtlineare Korrelationen zwi¬ schen den Abtastwerten der Zeitreihe, in diesem Fall des Elektrokardiogramm-Signals, vorhanden sind.FIG. 2 shows a further development of the method according to the invention, which considerably accelerates the feasibility of the method from measuring the time series to classifying the time series. This is achieved through a simplified classification. The simplification is that only on the basis of the parameterized dynamic diversity is it checked whether the samples of the time series are statistically dependent on one another 201. If this is the case, the time series is classified 202 as a first time series type 202. This classification corresponds for the application example of the classification of electrocardiogram signals in endangered and non-endangered hearts with regard to the sudden cardiac death of a classification of the electrocardiogram signal as a signal of an endangered heart, since in this case non-linear correlations between the samples of the time series, in in this case of the electrocardiogram signal.
Sind die Abtastwerte der Zeitreihe jedoch statistisch unab¬ hängig, wird die Zeitreihe als ein zweiter Zeitreihentyp klassifiziert 203. Dies entspricht in dem oben beschriebenen Beispiel für das Elektrokardiogramm-Signal einer Klassifika¬ tion als ein Signal eines bezüglich des Plötzlichen Herztodes nicht gefährdeten Herzens.However, if the samples of the time series are statistically independent, the time series is classified 203 as a second time series type. In the example described above for the electrocardiogram signal, this corresponds to a classification as a signal of a heart that is not at risk with regard to sudden cardiac death.
In Figur 3 sind in einer nicht abschließend zu verstehenden Übersicht einige Beispiele für mögliche Arten von Zeitreihen angegeben, für die das erfindungsgemäße Verfahren eingesetzt werden kann 301:In FIG. 3, an overview that cannot be conclusively understood gives some examples of possible types of time series for which the method according to the invention can be used 301:
- Elektrokardiogramm-Signale (EKG-Signale) 302; - Elektroencephalogramm-Signale (EEG-Signale) 303;Electrocardiogram signals (EKG signals) 302; Electroencephalogram signals (EEG signals) 303;
- Signale, die den Verlauf der lokalen SauerstoffSpannung eines Gehirns beschreiben 304.- Signals that describe the course of the local oxygen tension of a brain 304.
Eine Möglichkeit, ein Signal zu messen, das den Verlauf der lokalen SauerstoffSpannung in einem Gehirn beschreibt, ist in [2] dargestellt.One way to measure a signal that describes the course of the local oxygen voltage in a brain is shown in [2].
Das Verfahren kann selbstverständlich in allen Gebieten, in denen es gilt, anhand von parametrisierten dynamischen Man¬ nigfaltigkeiten eine Zeitreihe zu klassifizieren, eingesetzt werden.The method can of course be used in all areas in which it is important to classify a time series on the basis of parameterized dynamic manifolds.
In Figur 4 ist der Rechner R dargestellt, mit dem das erfin- dungsgemäße Verfahren notwendigerweise durchgeführt wird.FIG. 4 shows the computer R with which the method according to the invention is necessarily carried out.
Der Rechner R verarbeitet die von dem Meßgerät MG aufgenomme¬ nen, und dem Rechner R zugeführten Zeitreihen.The computer R processes the time series recorded by the measuring device MG and fed to the computer R.
Hierbei ist es nicht von Bedeutung, ob die Bildung der Ab¬ tastwerte aus dem möglicherweise analogen Signal in dem Me߬ gerät MG oder in dem Rechner R durchgeführt wird. Beide Vari¬ anten sind für das erfindungsgemäße Verfahren vorgesehen.It is not important here whether the formation of the sampled values from the possibly analog signal is carried out in the measuring device MG or in the computer R. Both variants are provided for the method according to the invention.
Das Meßgerät MG kann beispielsweise ein Elektrokardiograph (EKG) , ein Elektroencepahlograph (EEG) oder auch ein Gerät sein, welches nach dem in [2] dargestellten Verfahren arbei¬ tet.The measuring device MG can be, for example, an electrocardiograph (EKG), an electroencepahlograph (EEG) or also a device which works according to the method shown in [2].
Das Klassifikationsergebnis, welches durch den Rechner R auf die im vorigen beschriebene Weise ermittelt wird, wird in ei¬ nem Mittel zur Weiterverarbeitung WV weiterverarbeitet, bei¬ spielsweise einem Benutzer dargestellt. Dieses Mittel WV kann beispielsweise ein Drucker, ein Bildschirm oder auch ein Lautsprecher sein, über das ein akustisches oder visuelles Signal an einen Benutzer weitergegeben wird. Im Rahmen dieser Erfindung wurden folgende Veröffentlichungen zitiert:The classification result, which is ascertained by the computer R in the manner described above, is further processed in a means for further processing WV, for example presented to a user. This means WV can be, for example, a printer, a screen or a loudspeaker, via which an acoustic or visual signal is passed on to a user. The following publications were cited in the context of this invention:
[1] G. Morfill, Komplexitätsanalyse in der Kardiologie, Physikalische Blätter, Vol. 50, Nr. 2, S. 156 bis 160, 1994[1] G. Morfill, Complexity Analysis in Cardiology, Physikalische Blätter, Vol. 50, No. 2, pp. 156 to 160, 1994
[2] LICOX, GMS, Gesellschaft für Medizinische Sondentechnik mbH, Advanced Tissue Monitoring[2] LICOX, GMS, Society for Medical Sondentechnik mbH, Advanced Tissue Monitoring
[3] G. Deco & D. Schürmann, Learning Time Series Evolution by Unsupervised Extraction of Correlations, Physical Revue E, Vol. 51, Nr. 3, S. 1780 bis 1790, März 1995 [3] G. Deco & D. Schürmann, Learning Time Series Evolution by Unsupervised Extraction of Correlations, Physical Revue E, Vol. 51, No. 3, pp. 1780 to 1790, March 1995

Claims

Patentansprüche claims
1. Verfahren zur Klassifikation einer Zeitreihe, die eine vorgebbare Anzahl von Abtastwerten aufweist, insbesondere ei- ' nes elektrischen Signals, durch einen Rechner1. A method for classifying a time series which has a predeterminable number of samples, in particular an electrical signal, by a computer
- bei dem aus den Abtastwerten eine parametrisierte dynami¬ sche Mannigfaltigkeit ermittelt wird (102) , mit der nichtli¬ neare Korrelationen zwischen den Abtastwerten der Zeitreihe beschrieben werden, und - bei dem anhand der parametrisierten dynamischen Mannigfal¬ tigkeit eine Klassifikation der Zeitreihe durchgeführt wird (103) .- in which a parameterized dynamic manifold is determined (102) from the sampled values, with which nonlinear correlations between the sampled values of the time series are described, and - in which a classification of the time series is carried out on the basis of the parameterized dynamic manifold ( 103).
2. Verfahren nach Anspruch 1, bei dem bei der Klassifikation die Zeitreihe entweder in ei¬ nen ersten Zeitreihentyp oder in einen zweiten Zeitreihentyp klassifiziert wird.2. The method as claimed in claim 1, in which, during the classification, the time series is classified either into a first time series type or into a second time series type.
3. Verfahren nach Anspruch 1 oder 2, bei dem die Zeitreihe ein Elektrokardiogramm-Signal (EKG- Signal) ist.3. The method of claim 1 or 2, wherein the time series is an electrocardiogram signal (EKG signal).
4. Verfahren nach Anspruch 1 oder 2, bei dem die Zeitreihe ein Elektroencephalogramm-Signal (EEG- Signal) ist.4. The method of claim 1 or 2, wherein the time series is an electroencephalogram signal (EEG signal).
5. Verfahren nach Anspruch 1 oder 2, bei dem die Zeitreihe ein Signal ist, mit dem der zeitliche Verlauf einer Lokalen SauerstoffSpannung eines Gehirns (tip02j beschrieben wird. 5. The method of claim 1 or 2, wherein the time series is a signal with which the temporal course of a local oxygen voltage of a brain (tip02j is described.
EP97914158A 1996-03-06 1997-02-27 Process for classification of a time series which contains a pre-set number of sample values, particularly of an electrical signal, by means of a computer Withdrawn EP0885422A1 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
DE19608733 1996-03-06
DE19608733A DE19608733C1 (en) 1996-03-06 1996-03-06 Classification of time series with sampling values for evaluating EEG or ECG
PCT/DE1997/000350 WO1997033238A1 (en) 1996-03-06 1997-02-27 Process for classification of a time series which contains a pre-set number of sample values, particularly of an electrical signal, by means of a computer

Publications (1)

Publication Number Publication Date
EP0885422A1 true EP0885422A1 (en) 1998-12-23

Family

ID=7787441

Family Applications (1)

Application Number Title Priority Date Filing Date
EP97914158A Withdrawn EP0885422A1 (en) 1996-03-06 1997-02-27 Process for classification of a time series which contains a pre-set number of sample values, particularly of an electrical signal, by means of a computer

Country Status (5)

Country Link
US (1) US6117074A (en)
EP (1) EP0885422A1 (en)
JP (1) JP2000506039A (en)
DE (1) DE19608733C1 (en)
WO (1) WO1997033238A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
RU2751816C1 (en) * 2020-08-17 2021-07-19 Автономная некоммерческая организация высшего образования «Университет Иннополис» Method for classification of human motor activity

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE19610847C1 (en) * 1996-03-19 1997-04-30 Siemens Ag Electrocardiogram or electroencephalogram time series measurement signal classification system
DE19826993B4 (en) * 1998-06-19 2004-11-04 Forschungszentrum Jülich GmbH Imaging method for magnetic resonance signals
DE10036842A1 (en) * 2000-07-28 2002-02-07 Biotronik Mess & Therapieg Evaluation method for determining the temporal stationarity of measured physiological signals
US6718199B2 (en) * 2000-10-27 2004-04-06 Massachusetts Eye & Ear Infirmary Measurement of electrophysiologic response
JP2002358397A (en) * 2001-06-04 2002-12-13 Shigeru Suganuma Management method for time series data, information disclosing device and recording means
US8805482B2 (en) * 2008-07-28 2014-08-12 General Electric Conpany System and method for signal quality indication and false alarm reduction in ECG monitoring systems
DE102014101814A1 (en) * 2014-02-13 2015-08-13 Arthur Schultz Method for automatic evaluation of a Absence EEG, computer program and evaluation device therefor
CN111684540B (en) * 2017-12-08 2024-06-07 深圳迈瑞生物医疗电子股份有限公司 Statistical display method of physiological parameters in monitoring equipment and monitoring equipment

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4893632A (en) * 1988-04-13 1990-01-16 Siemens Aktiengesellschaft Method and apparatus for comparing waveform shapes of time-varying signals
US5792062A (en) * 1996-05-14 1998-08-11 Massachusetts Institute Of Technology Method and apparatus for detecting nonlinearity in an electrocardiographic signal

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See references of WO9733238A1 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
RU2751816C1 (en) * 2020-08-17 2021-07-19 Автономная некоммерческая организация высшего образования «Университет Иннополис» Method for classification of human motor activity

Also Published As

Publication number Publication date
US6117074A (en) 2000-09-12
WO1997033238A1 (en) 1997-09-12
DE19608733C1 (en) 1997-05-22
JP2000506039A (en) 2000-05-23

Similar Documents

Publication Publication Date Title
DE69624797T2 (en) METHOD AND APPARATUS FOR DETECTING ARTIFACTS USING EQUAL-SIGNAL SIGNALS IN DIFFERENTIAL SIGNAL DETECTORS
EP2704629B1 (en) Method for consciousness and pain monitoring, module for analyzing eeg signals and eeg narcosis monitor
DE69727236T2 (en) DEVICE FOR RECORDING GALVANIC SKIN REACTIONS
DE69722510T2 (en) REDUCTION OF INTERFERENCES IN ELEMENTARY MASS SPECTROMETRY
EP0828225A1 (en) Process and means for analysing EEG data
EP0355506A1 (en) Arrangement for measuring local bioelectric currents in biological tissue
DE3912028A1 (en) METHOD AND ARRANGEMENT FOR COMPARING WAVEFORMS OF TIME-VARIABLE SIGNALS
WO1999044498A1 (en) Evaluation system for obtaining diagnostic information from the signals and data of medical sensor systems
DE10042342B4 (en) Method and device for automatic detection and interpretation of pacemaker-influenced electrocardiograms
DE202022103434U1 (en) An encephalogram (EEG)-based device used to detect epileptic seizures
EP0885422A1 (en) Process for classification of a time series which contains a pre-set number of sample values, particularly of an electrical signal, by means of a computer
EP3338297B1 (en) Method for analysing a data set of a time-of-flight mass spectrometry measurement, and a device
EP1519301B1 (en) Apparatus for classifying physiological events
DE4327429C2 (en) Method and device for brain wave analysis
DE1964616C2 (en) Arrangement for the reproduction of cardiograms
DE19610847C1 (en) Electrocardiogram or electroencephalogram time series measurement signal classification system
DE19830316B4 (en) Method and apparatus for determining the likelihood of ventricular fibrillation
DE69732757T2 (en) Procedure and installation for QRS-Dektektion
DE10393475T5 (en) Method and device for identifying compounds in a sample
EP1192897B1 (en) Risk monitoring
EP0978053A1 (en) Process for classifying a time series having a predeterminable number of scanning values, e.g. an electric signal, by means of a computer
EP0885423B1 (en) Method of and apparatus for classification of a time series of an electrical signal by a computer,
DE102010024965A1 (en) Method for determining a characteristic course of an R-wave in an ECG signal, computer program product and electronically readable data carrier for carrying out the method
DE102010025920B4 (en) A method for generating a trigger signal by means of an ECG signal, in particular an R wave of the ECG signal, as well as an ECG measuring device and a magnetic resonance device
EP1519303A2 (en) Device for classifying physiological events

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 19980819

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): DE FR GB

17Q First examination report despatched

Effective date: 19990416

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN

18D Application deemed to be withdrawn

Effective date: 20030902